The purpose of this study was to develop linked risk prediction models for potential donors for and their respective recipients of a kidney transplantation at the time of counseling. We used characteristics available at the time of counseling from first single-organ transplantations performed at the Oslo University Hospital from 1854 donors between 1980 and 2007 with a median follow-up of 14.6 years, and from 837 recipients between 1995 and 2007 with a median follow-up of 13.1 years to derive multivariable Cox models to predict donor and recipient mortality, and graft loss. Deaths occurred in 195 donors and 255 recipients, and 162 grafts failed during the observation period until March 2015. Missing data were replaced by multiple imputation with chained equations. Statistical analyses were done using R and SAS. Variables were selected using multivariable fractional polynomials optimizing Akaike's information criterion. Age, calendar year of donation, smoking status, cholesterol and creatinine were selected and predicted donor mortality with a c-index of 0.821. Linear predictors from the donor mortality model served as parsimonious summary of donor prognosis and were considered among the candidate predictors for developing recipient models. Recipient age, recipient gender, calendar year of transplantation, pre-transplant dialysis vintage, primary renal disease, cerebrovascular disease, peripheral vascular disease, and HLA class II mismatch were selected and predict recipient mortality with a c-index of 0.763. Recipient age, pre-transplant dialysis vintage, linear predictor of donor risk model, HLA class I and II mismatch, peripheral vascular disease, and heart disease were selected and predict graft loss with a c-index of 0.682. We were able to demonstrate that donor survival prognosis is an important predictor of graft survival and present our linked risk prediction models derived from a well-maintained national database to support the decision making process for the recipient and his or her potential living kidney donor.

CITATION INFORMATION: Haller M, Mjøen G, Holdaas H, Heinze G, Oberbauer R. Risk Prediction Models for Living Donor Kidney Transplantation to Predict Donor and Recipient Mortality and Graft Loss. Am J Transplant. 2017;17 (suppl 3).

« Back to 2017 American Transplant Congress